During the operation of the electrical power system, power failure due to the distribution network faults statistically accounts for over 95% of the total accidents, and 70% of the accidents are caused by single-phase grounding faults or busbar faults. To avoid power failure when the single-phase grounding fault occurs, non-effective grounding (low-current grounding) mode at the neutral points of the external distribution network has been extensively used worldwide. The faulty line and location of the single-phase grounding fault on the distribution network has been difficult to detect, as the characteristics of the fault is weak. As the distribution network grows more automated, the fundamental need to solve the problem of locating the distribution network fault grows higher.
The current fault detection methods are generally divided into two categories, namely, the signal injection method and the section location method based on the faulty feature vector. The signal injection method includes “S” injection method, AC/DC injection method, and resistance in parallel method; the methods may impose high interference to the system and can't detect instantaneous and intermittent grounding faults. The section location method based on the faulty feature vector includes a zero-module current comparison method, section zero sequence admittance method, zero-sequence reaction power direction method, location method based on sudden difference of the phase current, residual increment method, and travel wave method; the methods have the disadvantages of difficulty to obtain the feature vector and precise synchronization of signals. Furthermore, some methods are not reliable when a high resistance is grounded.